Brown University’s Research Uncovers How Brains Encode Dynamic Rhythms
Summarize this article with:
Researchers at Brown University’s Carney Institute for Brain Science have created an artificial neural network offering new insight into how brains encode complex, rhythmic behaviors, specifically demonstrating how a four-legged animal might seamlessly transition between different gaits. Published in Neural Computation, the model utilizes a streamlined network of 24 artificial neurons to generate five distinct quadruped movements, bounding, pacing, trotting, walking, and pronking, and rapidly switch between them without parameter adjustments. “We know the brain has to be able to flexibly and robustly maintain and change rhythms,” said Carina Curto, a professor of applied mathematics at Brown. This work expands the established “attractor network” framework, potentially offering a more flexible and interpretable model for understanding brain function and inspiring more autonomous movement in quadruped robots.
Attractor Networks Model Dynamic Quadruped Gait Transitions This achievement challenges previous assumptions about the complexity required to model dynamic animal movement and provides a potential pathway toward more agile and autonomous robots.
The team’s model builds upon existing “attractor network” theory, a mathematical framework explaining how neural activity settles into specific patterns; traditional Hopfield networks model static behaviors like memory recall, but this research extends the concept to encompass dynamic actions. According to lead author Juliana Londono Alvarez, “This paper shows that you can expand attractor networks beyond the static to include the dynamic,” suggesting a unifying principle underlying both memory and movement. The efficiency of the network, its ability to generate complex behavior with minimal computational resources, is particularly noteworthy, as it contrasts sharply with the bulky, internet-dependent programming currently used in many quadruped robots. This streamlined approach could allow for the development of robots capable of operating offline and adapting to unpredictable terrain with greater ease; Londono Alvarez is already exploring collaborations with roboticists to implement the network in real-world applications. The research, published in Neural Computation, was supported by grants from the National Institutes of Health and the National Science Foundation, and benefited from a collaborative residency at Brown’s ICERM mathematics institute, emphasizing the interdisciplinary nature of this innovative work.
Neuron Network Generates Five Distinct Gaits Offline The network, comprised of only 24 artificial neurons, achieves these transitions, from a sudden leap to a shift between a trot and walk, without requiring any adjustments to its internal parameters, a level of efficiency previously unseen in comparable models. This streamlined approach contrasts sharply with current quadruped robot programming, which often relies on extensive, internet-dependent systems; the Brown team’s network, operating entirely offline, presents a pathway toward more autonomous robotic locomotion.
The team’s success in extending the attractor framework is significant, as it suggests a unified theoretical basis for understanding a broader range of brain functions, moving beyond simple static behaviors. This paper shows that you can expand attractor networks beyond the static to include the dynamic.
Juliana Londono Alvarez, a postdoctoral researcher at Brown and the study’s lead author Source: https://www.brown.edu/news/2026-03-23/ai-animal-gaits Tags: Quantum News There is so much happening right now in the field of technology, whether AI or the march of robots. Adrian is an expert on how technology can be transformative, especially frontier technologies. But Quantum occupies a special space. Quite literally a special space. A Hilbert space infact, haha! Here I try to provide some of the news that is considered breaking news in the Quantum Computing and Quantum tech space. Latest Posts by Quantum News: NVIDIA Builds Framework to Accelerate Simulation Data for AI March 24, 2026 Anthropic Explores How AI is Accelerating Pace of Scientific Discovery March 24, 2026 Anthropic Demonstrates AI’s Capacity for Frontier Theoretical Physics March 24, 2026